@openclaw/baseten-provider 2026.7.2-beta.2

package/README.md

@@ -0,0 +1,12 @@
+# OpenClaw Baseten Provider
+
+Official OpenClaw provider plugin for Baseten Model APIs, including Thinking Machines Lab's Inkling.
+
+Install from OpenClaw:
+
+```bash
+openclaw plugins install @openclaw/baseten-provider
+openclaw gateway restart
+```
+
+See <https://docs.openclaw.ai/providers/baseten> for setup and configuration.

package/dist/api.js

@@ -0,0 +1,4 @@
+import { a as buildBasetenModelCompat, c as discoverBasetenModels, d as usesBasetenChatTemplateThinking, i as BASETEN_MODEL_CATALOG, l as projectBasetenLiveModels, n as BASETEN_DEFAULT_MODEL_ID, o as buildBasetenModelDefinition, r as BASETEN_DEFAULT_MODEL_REF, s as buildStaticBasetenModels, t as BASETEN_BASE_URL, u as resolveBasetenDynamicModel } from "./models-Dg5jGc7M.js";
+import { applyBasetenConfig } from "./onboard.js";
+import { buildBasetenProvider, buildStaticBasetenProvider } from "./provider-catalog.js";
+export { BASETEN_BASE_URL, BASETEN_DEFAULT_MODEL_ID, BASETEN_DEFAULT_MODEL_REF, BASETEN_MODEL_CATALOG, applyBasetenConfig, buildBasetenModelCompat, buildBasetenModelDefinition, buildBasetenProvider, buildStaticBasetenModels, buildStaticBasetenProvider, discoverBasetenModels, projectBasetenLiveModels, resolveBasetenDynamicModel, usesBasetenChatTemplateThinking };

package/dist/index.js

@@ -0,0 +1,57 @@
+import { r as BASETEN_DEFAULT_MODEL_REF, u as resolveBasetenDynamicModel } from "./models-Dg5jGc7M.js";
+import { applyBasetenConfig } from "./onboard.js";
+import { buildBasetenProvider, buildStaticBasetenProvider } from "./provider-catalog.js";
+import { createBasetenThinkingWrapper } from "./stream.js";
+import { resolveBasetenThinkingProfile } from "./thinking.js";
+import { defineSingleProviderPluginEntry } from "openclaw/plugin-sdk/provider-entry";
+import { buildProviderReplayFamilyHooks } from "openclaw/plugin-sdk/provider-model-shared";
+//#region extensions/baseten/index.ts
+const PROVIDER_ID = "baseten";
+var baseten_default = defineSingleProviderPluginEntry({
+	id: PROVIDER_ID,
+	name: "Baseten Provider",
+	description: "Official Baseten Model APIs provider plugin",
+	provider: {
+		label: "Baseten",
+		docsPath: "/providers/baseten",
+		auth: [{
+			methodId: "api-key",
+			label: "Baseten API key",
+			hint: "Hosted Model APIs, including Inkling",
+			optionKey: "basetenApiKey",
+			flagName: "--baseten-api-key",
+			envVar: "BASETEN_API_KEY",
+			promptMessage: "Enter Baseten API key",
+			defaultModel: BASETEN_DEFAULT_MODEL_REF,
+			applyConfig: (cfg) => applyBasetenConfig(cfg),
+			noteTitle: "Baseten",
+			noteMessage: ["Baseten hosts Thinking Machines Lab's Inkling and other frontier models behind one OpenAI-compatible API.", "Get your API key at: https://app.baseten.co/settings/api_keys"].join("\n"),
+			wizard: {
+				groupLabel: "Baseten",
+				groupHint: "Hosted Model APIs, including Inkling"
+			}
+		}],
+		catalog: {
+			order: "simple",
+			run: async (ctx) => {
+				const { apiKey, discoveryApiKey } = ctx.resolveProviderAuth(PROVIDER_ID);
+				if (!apiKey) return null;
+				return { provider: {
+					...await buildBasetenProvider(discoveryApiKey),
+					apiKey
+				} };
+			},
+			staticRun: async () => ({ provider: buildStaticBasetenProvider() })
+		},
+		resolveDynamicModel: ({ modelId }) => resolveBasetenDynamicModel(modelId),
+		...buildProviderReplayFamilyHooks({
+			family: "openai-compatible",
+			dropReasoningFromHistory: false
+		}),
+		wrapStreamFn: (ctx) => createBasetenThinkingWrapper(ctx),
+		resolveThinkingProfile: ({ modelId }) => resolveBasetenThinkingProfile(modelId),
+		isModernModelRef: () => true
+	}
+});
+//#endregion
+export { baseten_default as default };

package/dist/models-Dg5jGc7M.js

@@ -0,0 +1,414 @@
+import { getCachedLiveProviderModelRows } from "openclaw/plugin-sdk/provider-catalog-live-runtime";
+import { buildManifestModelProviderConfig } from "openclaw/plugin-sdk/provider-catalog-shared";
+import { createSubsystemLogger } from "openclaw/plugin-sdk/runtime-env";
+import { ssrfPolicyFromHttpBaseUrlAllowedHostname } from "openclaw/plugin-sdk/ssrf-runtime";
+//#region extensions/baseten/openclaw.plugin.json
+var modelCatalog = {
+	"providers": { "baseten": {
+		"baseUrl": "https://inference.baseten.co/v1",
+		"api": "openai-completions",
+		"models": [
+			{
+				"id": "deepseek-ai/DeepSeek-V4-Pro",
+				"name": "DeepSeek V4 Pro",
+				"reasoning": true,
+				"input": ["text"],
+				"contextWindow": 262e3,
+				"maxTokens": 262e3,
+				"cost": {
+					"input": 1.74,
+					"output": 3.48,
+					"cacheRead": .145,
+					"cacheWrite": 0
+				}
+			},
+			{
+				"id": "zai-org/GLM-4.7",
+				"name": "GLM 4.7",
+				"reasoning": true,
+				"input": ["text"],
+				"contextWindow": 2e5,
+				"maxTokens": 2e5,
+				"cost": {
+					"input": .6,
+					"output": 2.2,
+					"cacheRead": .12,
+					"cacheWrite": 0
+				}
+			},
+			{
+				"id": "zai-org/GLM-5",
+				"name": "GLM 5",
+				"reasoning": true,
+				"input": ["text"],
+				"contextWindow": 202e3,
+				"maxTokens": 202e3,
+				"cost": {
+					"input": .95,
+					"output": 3.15,
+					"cacheRead": .2,
+					"cacheWrite": 0
+				}
+			},
+			{
+				"id": "zai-org/GLM-5.1",
+				"name": "GLM 5.1",
+				"reasoning": true,
+				"input": ["text"],
+				"contextWindow": 202e3,
+				"maxTokens": 202e3,
+				"cost": {
+					"input": 1.3,
+					"output": 4.3,
+					"cacheRead": .26,
+					"cacheWrite": 0
+				}
+			},
+			{
+				"id": "zai-org/GLM-5.2",
+				"name": "GLM 5.2",
+				"reasoning": true,
+				"input": ["text"],
+				"contextWindow": 202e3,
+				"maxTokens": 202e3,
+				"cost": {
+					"input": 1.4,
+					"output": 4.4,
+					"cacheRead": .26,
+					"cacheWrite": 0
+				}
+			},
+			{
+				"id": "thinkingmachines/inkling",
+				"name": "Inkling",
+				"reasoning": true,
+				"input": ["text", "image"],
+				"contextWindow": 1048e3,
+				"maxTokens": 32e3,
+				"cost": {
+					"input": 1,
+					"output": 4.05,
+					"cacheRead": .17,
+					"cacheWrite": 0
+				}
+			},
+			{
+				"id": "moonshotai/Kimi-K2.5",
+				"name": "Kimi K2.5",
+				"reasoning": true,
+				"input": ["text", "image"],
+				"contextWindow": 262e3,
+				"maxTokens": 262e3,
+				"cost": {
+					"input": .6,
+					"output": 3,
+					"cacheRead": .12,
+					"cacheWrite": 0
+				}
+			},
+			{
+				"id": "moonshotai/Kimi-K2.6",
+				"name": "Kimi K2.6",
+				"reasoning": true,
+				"input": ["text", "image"],
+				"contextWindow": 262e3,
+				"maxTokens": 262e3,
+				"cost": {
+					"input": .95,
+					"output": 4,
+					"cacheRead": .16,
+					"cacheWrite": 0
+				}
+			},
+			{
+				"id": "moonshotai/Kimi-K2.7-Code",
+				"name": "Kimi K2.7 Code",
+				"reasoning": true,
+				"input": ["text", "image"],
+				"contextWindow": 262e3,
+				"maxTokens": 262e3,
+				"cost": {
+					"input": .95,
+					"output": 4,
+					"cacheRead": .16,
+					"cacheWrite": 0
+				}
+			},
+			{
+				"id": "nvidia/Nemotron-120B-A12B",
+				"name": "Nemotron Super",
+				"reasoning": true,
+				"input": ["text"],
+				"contextWindow": 202e3,
+				"maxTokens": 202e3,
+				"cost": {
+					"input": .3,
+					"output": .75,
+					"cacheRead": .06,
+					"cacheWrite": 0
+				}
+			},
+			{
+				"id": "nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B",
+				"name": "Nemotron Ultra",
+				"reasoning": true,
+				"input": ["text"],
+				"contextWindow": 202e3,
+				"maxTokens": 202e3,
+				"cost": {
+					"input": .6,
+					"output": 2.4,
+					"cacheRead": .12,
+					"cacheWrite": 0
+				}
+			},
+			{
+				"id": "openai/gpt-oss-120b",
+				"name": "GPT OSS 120B",
+				"reasoning": true,
+				"input": ["text"],
+				"contextWindow": 128e3,
+				"maxTokens": 128e3,
+				"cost": {
+					"input": .1,
+					"output": .5,
+					"cacheRead": 0,
+					"cacheWrite": 0
+				}
+			}
+		]
+	} },
+	"discovery": { "baseten": "refreshable" }
+};
+//#endregion
+//#region extensions/baseten/models.ts
+/**
+* Baseten model catalog, compat metadata, and authenticated live discovery.
+*/
+const log = createSubsystemLogger("baseten-models");
+const BASETEN_MANIFEST_CATALOG = modelCatalog.providers.baseten;
+const CACHE_TTL_MS = 300 * 1e3;
+const DEFAULT_CONTEXT_WINDOW = 128e3;
+const DEFAULT_MAX_TOKENS = 8192;
+const CHAT_TEMPLATE_THINKING_MODEL_IDS = /* @__PURE__ */ new Set([
+	"zai-org/glm-4.7",
+	"zai-org/glm-5",
+	"zai-org/glm-5.1",
+	"zai-org/glm-5.2",
+	"moonshotai/kimi-k2.5",
+	"moonshotai/kimi-k2.6",
+	"moonshotai/kimi-k2.7-code",
+	"nvidia/nemotron-120b-a12b",
+	"nvidia/nvidia-nemotron-3-ultra-550b-a55b"
+]);
+const FULL_REASONING_EFFORT_MODEL_IDS = /* @__PURE__ */ new Set(["deepseek-ai/DeepSeek-V4-Pro", "openai/gpt-oss-120b"]);
+const INKLING_REASONING_EFFORTS = [
+	"none",
+	"minimal",
+	"low",
+	"medium",
+	"high",
+	"xhigh"
+];
+const FULL_REASONING_EFFORTS = [...INKLING_REASONING_EFFORTS, "max"];
+const BASE_COMPAT = {
+	supportsStore: false,
+	supportsDeveloperRole: false,
+	supportsUsageInStreaming: true,
+	supportsStrictMode: true,
+	supportsTools: true,
+	maxTokensField: "max_tokens"
+};
+/** Base URL for Baseten's OpenAI-compatible Model APIs. */
+const BASETEN_BASE_URL = BASETEN_MANIFEST_CATALOG.baseUrl;
+/** Default Baseten model id used for onboarding. */
+const BASETEN_DEFAULT_MODEL_ID = "thinkingmachines/inkling";
+/** Default Baseten model ref used for onboarding. */
+const BASETEN_DEFAULT_MODEL_REF = `baseten/${BASETEN_DEFAULT_MODEL_ID}`;
+/** Bundled fallback rows for all Baseten Model APIs available at release time. */
+const BASETEN_MODEL_CATALOG = BASETEN_MANIFEST_CATALOG.models;
+/** Whether Baseten requires chat-template thinking control for this model. */
+function usesBasetenChatTemplateThinking(modelId) {
+	return CHAT_TEMPLATE_THINKING_MODEL_IDS.has(modelId.trim().toLowerCase());
+}
+function buildBasetenReasoningCompat(modelId) {
+	if (FULL_REASONING_EFFORT_MODEL_IDS.has(modelId)) return {
+		supportsReasoningEffort: true,
+		supportedReasoningEfforts: FULL_REASONING_EFFORTS,
+		reasoningEffortMap: {
+			off: "none",
+			none: "none",
+			adaptive: "max"
+		}
+	};
+	if (modelId === "thinkingmachines/inkling") return {
+		supportsReasoningEffort: true,
+		supportedReasoningEfforts: INKLING_REASONING_EFFORTS,
+		reasoningEffortMap: {
+			off: "none",
+			none: "none",
+			adaptive: "xhigh",
+			max: "xhigh"
+		}
+	};
+	if (modelId === "zai-org/GLM-5.2") return {
+		supportsReasoningEffort: true,
+		supportedReasoningEfforts: [
+			"none",
+			"high",
+			"max"
+		],
+		reasoningEffortMap: {
+			off: "none",
+			none: "none",
+			minimal: "high",
+			low: "high",
+			medium: "high",
+			xhigh: "high",
+			adaptive: "max"
+		}
+	};
+	return {};
+}
+/** Complete OpenAI-compatible transport policy for one Baseten model. */
+function buildBasetenModelCompat(modelId) {
+	return {
+		...BASE_COMPAT,
+		...buildBasetenReasoningCompat(modelId)
+	};
+}
+/** Builds one normalized Baseten model definition from a manifest entry. */
+function buildBasetenModelDefinition(model) {
+	const normalized = buildManifestModelProviderConfig({
+		providerId: "baseten",
+		catalog: {
+			...BASETEN_MANIFEST_CATALOG,
+			models: [model]
+		}
+	}).models[0];
+	if (!normalized) throw new Error(`Missing normalized Baseten model ${model.id}`);
+	return {
+		...normalized,
+		compat: buildBasetenModelCompat(normalized.id)
+	};
+}
+/** Builds the network-free fallback catalog. */
+function buildStaticBasetenModels() {
+	return BASETEN_MODEL_CATALOG.map(buildBasetenModelDefinition);
+}
+function readPositiveInteger(value) {
+	const number = typeof value === "number" ? value : Number(value);
+	return Number.isSafeInteger(number) && number > 0 ? number : void 0;
+}
+function readPerTokenPrice(value) {
+	if (typeof value !== "number" && (typeof value !== "string" || !value.trim())) return;
+	const number = typeof value === "number" ? value : Number(value);
+	return Number.isFinite(number) && number >= 0 ? Number((number * 1e6).toFixed(9)) : void 0;
+}
+function readStringArray(value) {
+	return Array.isArray(value) ? value.filter((item) => typeof item === "string") : [];
+}
+function applyLiveReasoningEffortCompat(fallbackCompat, supportsReasoningEffort) {
+	if (supportsReasoningEffort) return {
+		...fallbackCompat,
+		supportsReasoningEffort: true
+	};
+	const compat = { ...fallbackCompat };
+	delete compat.supportsReasoningEffort;
+	delete compat.supportedReasoningEfforts;
+	delete compat.reasoningEffortMap;
+	return compat;
+}
+function projectLiveModel(row, fallback) {
+	if (row.object !== void 0 && row.object !== "model") return;
+	const id = typeof row.id === "string" ? row.id.trim() : "";
+	if (!id) return;
+	const hasLiveFeatures = Array.isArray(row.supported_features);
+	const features = new Set(readStringArray(row.supported_features));
+	const pricing = row.pricing && typeof row.pricing === "object" && !Array.isArray(row.pricing) ? row.pricing : {};
+	const inputPrice = readPerTokenPrice(pricing.prompt);
+	const outputPrice = readPerTokenPrice(pricing.completion);
+	const cacheReadPrice = readPerTokenPrice(pricing.input_cache_read);
+	const supportsReasoningEffort = features.has("reasoning_effort");
+	const fallbackCompat = fallback?.compat ?? buildBasetenModelCompat(id);
+	const compat = hasLiveFeatures ? applyLiveReasoningEffortCompat(fallbackCompat, supportsReasoningEffort) : fallbackCompat;
+	return {
+		id,
+		name: typeof row.name === "string" && row.name.trim() ? row.name.trim() : fallback?.name ?? id,
+		reasoning: hasLiveFeatures ? features.has("reasoning") || supportsReasoningEffort : fallback?.reasoning ?? false,
+		input: hasLiveFeatures ? features.has("vision") ? ["text", "image"] : ["text"] : fallback?.input ?? ["text"],
+		cost: {
+			input: inputPrice ?? fallback?.cost.input ?? 0,
+			output: outputPrice ?? fallback?.cost.output ?? 0,
+			cacheRead: cacheReadPrice ?? fallback?.cost.cacheRead ?? 0,
+			cacheWrite: fallback?.cost.cacheWrite ?? 0
+		},
+		contextWindow: readPositiveInteger(row.context_length) ?? fallback?.contextWindow ?? DEFAULT_CONTEXT_WINDOW,
+		maxTokens: readPositiveInteger(row.max_completion_tokens) ?? fallback?.maxTokens ?? DEFAULT_MAX_TOKENS,
+		compat
+	};
+}
+/** Projects Baseten's authenticated `/models` response into OpenClaw model rows. */
+function projectBasetenLiveModels(rows) {
+	const fallbacks = new Map(buildStaticBasetenModels().map((model) => [model.id, model]));
+	const seen = /* @__PURE__ */ new Set();
+	const models = [];
+	for (const row of rows) {
+		if (!row || typeof row !== "object" || Array.isArray(row)) continue;
+		const model = projectLiveModel(row, fallbacks.get(String(row.id)));
+		if (!model || seen.has(model.id)) continue;
+		seen.add(model.id);
+		models.push(model);
+	}
+	return models;
+}
+/** Discovers every model enabled for a Baseten account, with a static fallback. */
+async function discoverBasetenModels(params = {}) {
+	const staticModels = buildStaticBasetenModels();
+	const env = params.env ?? process.env;
+	if (!params.discoveryApiKey?.trim() || !params.forceLive && (env.NODE_ENV === "test" || env.VITEST === "true")) return staticModels;
+	try {
+		const models = projectBasetenLiveModels(await getCachedLiveProviderModelRows({
+			providerId: "baseten",
+			endpoint: `${BASETEN_BASE_URL}/models`,
+			discoveryApiKey: params.discoveryApiKey,
+			fetchGuard: params.fetchGuard,
+			signal: params.signal,
+			timeoutMs: 1e4,
+			ttlMs: CACHE_TTL_MS,
+			policy: ssrfPolicyFromHttpBaseUrlAllowedHostname(BASETEN_BASE_URL),
+			auditContext: "baseten-model-discovery",
+			shouldCacheRows: (candidateRows) => projectBasetenLiveModels(candidateRows).length > 0
+		}));
+		if (models.length > 0) return models;
+		log.warn("Baseten returned no usable models; using bundled catalog");
+	} catch (error) {
+		log.warn(`Baseten model discovery failed; using bundled catalog: ${String(error)}`);
+	}
+	return staticModels;
+}
+/** Resolves a forward-compatible Baseten model id not yet in the bundled catalog. */
+function resolveBasetenDynamicModel(modelId) {
+	const id = modelId.trim();
+	if (!id || BASETEN_MODEL_CATALOG.some((model) => model.id === id)) return;
+	return {
+		id,
+		name: id,
+		provider: "baseten",
+		api: "openai-completions",
+		baseUrl: BASETEN_BASE_URL,
+		reasoning: false,
+		input: ["text"],
+		cost: {
+			input: 0,
+			output: 0,
+			cacheRead: 0,
+			cacheWrite: 0
+		},
+		contextWindow: DEFAULT_CONTEXT_WINDOW,
+		maxTokens: DEFAULT_MAX_TOKENS,
+		compat: buildBasetenModelCompat(id)
+	};
+}
+//#endregion
+export { buildBasetenModelCompat as a, discoverBasetenModels as c, usesBasetenChatTemplateThinking as d, BASETEN_MODEL_CATALOG as i, projectBasetenLiveModels as l, BASETEN_DEFAULT_MODEL_ID as n, buildBasetenModelDefinition as o, BASETEN_DEFAULT_MODEL_REF as r, buildStaticBasetenModels as s, BASETEN_BASE_URL as t, resolveBasetenDynamicModel as u };

package/dist/models.js

@@ -0,0 +1,2 @@
+import { a as buildBasetenModelCompat, c as discoverBasetenModels, d as usesBasetenChatTemplateThinking, i as BASETEN_MODEL_CATALOG, l as projectBasetenLiveModels, n as BASETEN_DEFAULT_MODEL_ID, o as buildBasetenModelDefinition, r as BASETEN_DEFAULT_MODEL_REF, s as buildStaticBasetenModels, t as BASETEN_BASE_URL, u as resolveBasetenDynamicModel } from "./models-Dg5jGc7M.js";
+export { BASETEN_BASE_URL, BASETEN_DEFAULT_MODEL_ID, BASETEN_DEFAULT_MODEL_REF, BASETEN_MODEL_CATALOG, buildBasetenModelCompat, buildBasetenModelDefinition, buildStaticBasetenModels, discoverBasetenModels, projectBasetenLiveModels, resolveBasetenDynamicModel, usesBasetenChatTemplateThinking };

package/dist/onboard.js

@@ -0,0 +1,23 @@
+import { r as BASETEN_DEFAULT_MODEL_REF, s as buildStaticBasetenModels, t as BASETEN_BASE_URL } from "./models-Dg5jGc7M.js";
+import { createModelCatalogPresetAppliers } from "openclaw/plugin-sdk/provider-onboard";
+//#region extensions/baseten/onboard.ts
+/** Baseten onboarding config helpers. */
+const basetenPresetAppliers = createModelCatalogPresetAppliers({
+	primaryModelRef: BASETEN_DEFAULT_MODEL_REF,
+	resolveParams: (_cfg) => ({
+		providerId: "baseten",
+		api: "openai-completions",
+		baseUrl: BASETEN_BASE_URL,
+		catalogModels: buildStaticBasetenModels(),
+		aliases: [{
+			modelRef: BASETEN_DEFAULT_MODEL_REF,
+			alias: "Inkling"
+		}]
+	})
+});
+/** Applies Baseten's provider catalog, Inkling alias, and default model. */
+function applyBasetenConfig(cfg) {
+	return basetenPresetAppliers.applyConfig(cfg);
+}
+//#endregion
+export { applyBasetenConfig };

package/dist/provider-catalog.js

@@ -0,0 +1,20 @@
+import { c as discoverBasetenModels, s as buildStaticBasetenModels, t as BASETEN_BASE_URL } from "./models-Dg5jGc7M.js";
+//#region extensions/baseten/provider-catalog.ts
+/** Builds Baseten's network-free fallback provider catalog. */
+function buildStaticBasetenProvider() {
+	return {
+		baseUrl: BASETEN_BASE_URL,
+		api: "openai-completions",
+		models: buildStaticBasetenModels()
+	};
+}
+/** Builds Baseten's account-scoped live catalog. */
+async function buildBasetenProvider(discoveryApiKey) {
+	return {
+		baseUrl: BASETEN_BASE_URL,
+		api: "openai-completions",
+		models: await discoverBasetenModels({ discoveryApiKey })
+	};
+}
+//#endregion
+export { buildBasetenProvider, buildStaticBasetenProvider };

package/dist/stream.js

@@ -0,0 +1,34 @@
+import { d as usesBasetenChatTemplateThinking } from "./models-Dg5jGc7M.js";
+import { createPayloadPatchStreamWrapper } from "openclaw/plugin-sdk/provider-stream-shared";
+//#region extensions/baseten/stream.ts
+const BASETEN_DEEPSEEK_V4_MODEL_ID = "deepseek-ai/deepseek-v4-pro";
+function isThinkingEnabled(level) {
+	return level !== void 0 && level !== "off";
+}
+function isBasetenDeepSeekV4ModelId(modelId) {
+	return modelId.trim().toLowerCase() === BASETEN_DEEPSEEK_V4_MODEL_ID;
+}
+function patchBasetenDeepSeekV4Replay(payload, thinkingEnabled) {
+	if (!Array.isArray(payload.messages)) return;
+	for (const message of payload.messages) {
+		if (!message || typeof message !== "object" || Array.isArray(message)) continue;
+		const record = message;
+		if (record.role !== "assistant") continue;
+		if (thinkingEnabled) record.reasoning_content ??= "";
+		else delete record.reasoning_content;
+	}
+}
+/** Adds Baseten's `chat_template_args.enable_thinking` without dropping caller args. */
+function createBasetenThinkingWrapper(ctx) {
+	return createPayloadPatchStreamWrapper(ctx.streamFn, ({ payload, model }) => {
+		if (model.provider !== "baseten" || model.api !== "openai-completions") return;
+		if (isBasetenDeepSeekV4ModelId(model.id)) patchBasetenDeepSeekV4Replay(payload, ctx.thinkingLevel !== "off");
+		if (!usesBasetenChatTemplateThinking(model.id)) return;
+		payload.chat_template_args = {
+			...payload.chat_template_args && typeof payload.chat_template_args === "object" && !Array.isArray(payload.chat_template_args) ? payload.chat_template_args : {},
+			enable_thinking: isThinkingEnabled(ctx.thinkingLevel)
+		};
+	});
+}
+//#endregion
+export { createBasetenThinkingWrapper };

package/dist/thinking.js

@@ -0,0 +1,25 @@
+import { d as usesBasetenChatTemplateThinking } from "./models-Dg5jGc7M.js";
+//#region extensions/baseten/thinking.ts
+const BASETEN_BINARY_THINKING_PROFILE = {
+	levels: [{ id: "off" }, {
+		id: "low",
+		label: "on"
+	}],
+	defaultLevel: "off"
+};
+const BASETEN_GLM_52_THINKING_PROFILE = {
+	levels: [
+		{ id: "off" },
+		{ id: "high" },
+		{ id: "max" }
+	],
+	defaultLevel: "off"
+};
+/** Exposes only the thinking levels that Baseten actually accepts for opt-in models. */
+function resolveBasetenThinkingProfile(modelId) {
+	const normalized = modelId.trim().toLowerCase();
+	if (normalized === "zai-org/glm-5.2") return BASETEN_GLM_52_THINKING_PROFILE;
+	return usesBasetenChatTemplateThinking(normalized) ? BASETEN_BINARY_THINKING_PROFILE : void 0;
+}
+//#endregion
+export { resolveBasetenThinkingProfile };

package/npm-shrinkwrap.json

@@ -0,0 +1,12 @@
+{
+  "name": "@openclaw/baseten-provider",
+  "version": "2026.7.2-beta.2",
+  "lockfileVersion": 3,
+  "requires": true,
+  "packages": {
+    "": {
+      "name": "@openclaw/baseten-provider",
+      "version": "2026.7.2-beta.2"
+    }
+  }
+}

package/openclaw.plugin.json

@@ -0,0 +1,230 @@
+{
+  "id": "baseten",
+  "name": "Baseten",
+  "description": "OpenClaw Baseten provider plugin.",
+  "activation": {
+    "onStartup": false
+  },
+  "enabledByDefault": true,
+  "providers": ["baseten"],
+  "providerRequest": {
+    "providers": {
+      "baseten": {
+        "family": "baseten",
+        "openAICompletions": {
+          "supportsStreamingUsage": true
+        }
+      }
+    }
+  },
+  "modelCatalog": {
+    "providers": {
+      "baseten": {
+        "baseUrl": "https://inference.baseten.co/v1",
+        "api": "openai-completions",
+        "models": [
+          {
+            "id": "deepseek-ai/DeepSeek-V4-Pro",
+            "name": "DeepSeek V4 Pro",
+            "reasoning": true,
+            "input": ["text"],
+            "contextWindow": 262000,
+            "maxTokens": 262000,
+            "cost": {
+              "input": 1.74,
+              "output": 3.48,
+              "cacheRead": 0.145,
+              "cacheWrite": 0
+            }
+          },
+          {
+            "id": "zai-org/GLM-4.7",
+            "name": "GLM 4.7",
+            "reasoning": true,
+            "input": ["text"],
+            "contextWindow": 200000,
+            "maxTokens": 200000,
+            "cost": {
+              "input": 0.6,
+              "output": 2.2,
+              "cacheRead": 0.12,
+              "cacheWrite": 0
+            }
+          },
+          {
+            "id": "zai-org/GLM-5",
+            "name": "GLM 5",
+            "reasoning": true,
+            "input": ["text"],
+            "contextWindow": 202000,
+            "maxTokens": 202000,
+            "cost": {
+              "input": 0.95,
+              "output": 3.15,
+              "cacheRead": 0.2,
+              "cacheWrite": 0
+            }
+          },
+          {
+            "id": "zai-org/GLM-5.1",
+            "name": "GLM 5.1",
+            "reasoning": true,
+            "input": ["text"],
+            "contextWindow": 202000,
+            "maxTokens": 202000,
+            "cost": {
+              "input": 1.3,
+              "output": 4.3,
+              "cacheRead": 0.26,
+              "cacheWrite": 0
+            }
+          },
+          {
+            "id": "zai-org/GLM-5.2",
+            "name": "GLM 5.2",
+            "reasoning": true,
+            "input": ["text"],
+            "contextWindow": 202000,
+            "maxTokens": 202000,
+            "cost": {
+              "input": 1.4,
+              "output": 4.4,
+              "cacheRead": 0.26,
+              "cacheWrite": 0
+            }
+          },
+          {
+            "id": "thinkingmachines/inkling",
+            "name": "Inkling",
+            "reasoning": true,
+            "input": ["text", "image"],
+            "contextWindow": 1048000,
+            "maxTokens": 32000,
+            "cost": {
+              "input": 1,
+              "output": 4.05,
+              "cacheRead": 0.17,
+              "cacheWrite": 0
+            }
+          },
+          {
+            "id": "moonshotai/Kimi-K2.5",
+            "name": "Kimi K2.5",
+            "reasoning": true,
+            "input": ["text", "image"],
+            "contextWindow": 262000,
+            "maxTokens": 262000,
+            "cost": {
+              "input": 0.6,
+              "output": 3,
+              "cacheRead": 0.12,
+              "cacheWrite": 0
+            }
+          },
+          {
+            "id": "moonshotai/Kimi-K2.6",
+            "name": "Kimi K2.6",
+            "reasoning": true,
+            "input": ["text", "image"],
+            "contextWindow": 262000,
+            "maxTokens": 262000,
+            "cost": {
+              "input": 0.95,
+              "output": 4,
+              "cacheRead": 0.16,
+              "cacheWrite": 0
+            }
+          },
+          {
+            "id": "moonshotai/Kimi-K2.7-Code",
+            "name": "Kimi K2.7 Code",
+            "reasoning": true,
+            "input": ["text", "image"],
+            "contextWindow": 262000,
+            "maxTokens": 262000,
+            "cost": {
+              "input": 0.95,
+              "output": 4,
+              "cacheRead": 0.16,
+              "cacheWrite": 0
+            }
+          },
+          {
+            "id": "nvidia/Nemotron-120B-A12B",
+            "name": "Nemotron Super",
+            "reasoning": true,
+            "input": ["text"],
+            "contextWindow": 202000,
+            "maxTokens": 202000,
+            "cost": {
+              "input": 0.3,
+              "output": 0.75,
+              "cacheRead": 0.06,
+              "cacheWrite": 0
+            }
+          },
+          {
+            "id": "nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B",
+            "name": "Nemotron Ultra",
+            "reasoning": true,
+            "input": ["text"],
+            "contextWindow": 202000,
+            "maxTokens": 202000,
+            "cost": {
+              "input": 0.6,
+              "output": 2.4,
+              "cacheRead": 0.12,
+              "cacheWrite": 0
+            }
+          },
+          {
+            "id": "openai/gpt-oss-120b",
+            "name": "GPT OSS 120B",
+            "reasoning": true,
+            "input": ["text"],
+            "contextWindow": 128000,
+            "maxTokens": 128000,
+            "cost": {
+              "input": 0.1,
+              "output": 0.5,
+              "cacheRead": 0,
+              "cacheWrite": 0
+            }
+          }
+        ]
+      }
+    },
+    "discovery": {
+      "baseten": "refreshable"
+    }
+  },
+  "setup": {
+    "providers": [
+      {
+        "id": "baseten",
+        "envVars": ["BASETEN_API_KEY"]
+      }
+    ]
+  },
+  "providerAuthChoices": [
+    {
+      "provider": "baseten",
+      "method": "api-key",
+      "choiceId": "baseten-api-key",
+      "appGuidedSecret": true,
+      "choiceLabel": "Baseten API key",
+      "groupId": "baseten",
+      "groupLabel": "Baseten",
+      "groupHint": "Hosted Model APIs, including Inkling",
+      "optionKey": "basetenApiKey",
+      "cliFlag": "--baseten-api-key",
+      "cliOption": "--baseten-api-key <key>",
+      "cliDescription": "Baseten API key"
+    }
+  ],
+  "configSchema": {
+    "type": "object",
+    "additionalProperties": false,
+    "properties": {}
+  }
+}

package/package.json

@@ -0,0 +1,50 @@
+{
+  "name": "@openclaw/baseten-provider",
+  "version": "2026.7.2-beta.2",
+  "description": "OpenClaw Baseten provider plugin.",
+  "repository": {
+    "type": "git",
+    "url": "https://github.com/openclaw/openclaw"
+  },
+  "type": "module",
+  "openclaw": {
+    "extensions": [
+      "./index.ts"
+    ],
+    "install": {
+      "clawhubSpec": "clawhub:@openclaw/baseten-provider",
+      "npmSpec": "@openclaw/baseten-provider",
+      "defaultChoice": "npm",
+      "minHostVersion": ">=2026.7.2-beta.2"
+    },
+    "compat": {
+      "pluginApi": ">=2026.7.2-beta.2"
+    },
+    "build": {
+      "openclawVersion": "2026.7.2-beta.2",
+      "bundledDist": false
+    },
+    "release": {
+      "publishToClawHub": true,
+      "publishToNpm": true
+    },
+    "runtimeExtensions": [
+      "./dist/index.js"
+    ]
+  },
+  "files": [
+    "dist/**",
+    "openclaw.plugin.json",
+    "npm-shrinkwrap.json",
+    "README.md"
+  ],
+  "peerDependencies": {
+    "openclaw": ">=2026.7.2-beta.2"
+  },
+  "peerDependenciesMeta": {
+    "openclaw": {
+      "optional": true
+    }
+  },
+  "bundledDependencies": []
+}